Three-screen delivery isn’t made possible by the type of science fiction King deals in, however. Sending signals via IP to a variety of devices is a complicated task. Arguably the most important enabler is adaptive streaming, and observers say all major MSOs are testing the technology.
“It’s an MSO top-of-agenda item,” says Anshu Agarwal, Juniper Networks’ [www.junipernetworks.com] head of product management and marketing for content and media. “With all cable operators, this is a thing that comes up. Today we are calling it ‘adaptive streaming.’ In a year or two it will be called just ‘streaming,’ because normal streaming will be adaptive.”
Dhaval Ajmera, EVP of sales and marketing at Verisimo Networks [www.verisimonetworks.com], agrees about the technique’s importance. “I think in the near future you will see a huge world of IPTV, and adaptive streaming will have a massive role,” he says.
It won’t be easy, however. “Depending on the geography, location, bit rate and bandwidth, [what is necessary for success] is very different,” Ajmera explains. “In the U.S., you can have 12 megabits to an end device within a metro, but on the outskirts you’ll find a drop off. Adaptive streaming is mandated in such situations.” Verisimo offers a platform for the delivery of video over the Internet and other IP networks.
Adaptive streaming is the ability to send the right amount of data to end users no matter what device they are using, even over unmanaged broadband networks. Precisely what that optimum throughput is depends on a combination of factors, including the nature of the end user device and the size and resolution of its display, the amount of bandwidth available for transmission for that user, the relative condition of that bandwidth – i.e., is it a noisy or pristine network -- and the nature of the programming for which the subscriber is asking.
Adaptive streaming is simple in concept. A specific piece of content generally is compressed into a variety of profiles, or versions, each of which operates at a different number of bits per second. End devices – iPads, Android smartphones, PCs or others – track how quickly segments of programming enter and leave their buffers, says Jeff Brooks, ARRIS’ [www.arrisi.com] VP for IP video product line management. If the bits comprising a piece of content are depleted faster than they are replaced, the device throttles up to a profile with a more aggressive rate of replenishment. Likewise, if too much content is being delivered for the device and network conditions at hand, the system downshifts to a more moderate profile – and frees up bandwidth in the bargain.
One of the big changes in streaming is that companies writing the adaptive streaming algorithms used by cable operators and other service providers have added many new profiles. This is, in essence, the empowering of the adaptive element. “The big difference is that there used to be three to five encoding profiles,” says Marty Roberts, VP of sales and marketing at thePlatform [www.theplatform.com], a company that offers the complex management and logistics for service providers offering online video.
Not only does adding profiles increase the higher end limits of picture quality, but it makes the incremental steps between each profile less noticeable to users. “There may have been a 300 kilobit, 500 kilobit, 750 kilobit and 1 megabit and 1.5 megabit profile,” Roberts says. “Now we are seeing that it starts around 300 kilobits per second and ranges up to 2 to 2.5 Mbps, and increases in many more steps. There are a lot more options to worth with there.”
A Work in Progress
Observers universally say that adaptive streaming, and cable operators’ use of it, still is very much a work in progress. Generally, content is compressed using the H.264 codec. Service providers can choose from three approaches: Microsoft’s Smooth Streaming -- part of Silverlight – Adobe’s Dynamic Streaming and Apple’s HTTP Live Streaming (HLS). While there is consensus on what the various profiles will deliver, there is none on the precise nature of the containers in which the segments of programming are carried through the network. In essence, the three do much the same thing, but do so differently enough to make them unable to interoperate.
Observers say that there are no meaningful steps being taken to create an industry-standardized approach. In other words, the industry may have to support a trio of protocols.
Nothing spells inefficiency as clearly as three competing standards. “You don’t want to force users to install a new player,” says Asaf Atzmon, the senior director in the office of the CTO for BigBand Networks [www.bigbandnetworks.com]. “We need a way to support existing protocols to use in the network in a different way … No one has shown an end-to-end full architecture that does that. At some point, the cable industry will come up with such an architecture that on one hand doesn’t require the user to change the client and in the other makes [efficient] use of the network.”
Being forced to support approaches by Apple, Adobe and Microsoft is “a big deal,” Andy Salo, director of product marketing for RGB Networks [www.rgbnetworks.com], explains. “It is not trivial … It is not a linear three-times amount of infrastructure, but arguably 1.5 or 2 times.” RGB provides transcoding infrastructure.
Clearly, the industry is at the beginning of a long process, and perhaps the emergence of IP video will lead to the creation of a more efficient and universal approach to implementation of adaptive streaming.
Cable, pushed by the explosion of mobility and the aggressiveness of its telco and satellite competitors, is reacting with significant testing of adaptive streaming, observers say. “Part of the trials is trying to find out which platforms and formats offer the best bang for the buck, which make the most sense and are easiest to implement in terms of digital rights management and security,” says Arnaud Perrier, the VP of solutions for Envivio [www.envivio.com], which offers a platform that enables the delivery of content in any of the three formats.
The cable industry, which is just getting its feet wet in the world of adaptive streaming, will deal with two related challenges. One is supporting end users accessing content aggregators – the Hulus and YouTubes of the world – in a purely over-the-top scenario. The other is optimizing delivery for services such as Comcast’s XFINITY and Rogers On Demand Online. In these cases, operators have a higher level of control over the infrastructure through which the subscriber is reached.
Jim Welch, senior consulting engineer at IneoQuest [www.ineoquest.com], thinks the cable industry has one advantage in its back pocket, especially when it is providing IP video over its own infrastructure. “In some respects, it will be a bit easier for MSOs because they may have the resources they need within their own network,” he says. “There could be less of a multivendor aggregation.” IneoQuest offers Cricket AdaptStream, a probe for adaptive streaming applications.
Cable operators’ need to support managed and unmanaged services raises questions that as much about business as technology. “From a solutions standpoint, the challenge is understanding where operators’ responsibility begins and where it ends,” says Sean Yarborough, senior director of strategy and business development for Spirent Communications [www.spirent.com]. “If a subscriber is using services of a company like Amazon or Netflix, what is the responsibility of the MSO? Should the operator offer a higher level QoS?”
The bottom line is that operators may do more for subscribers who are providing them more revenue. And, the industry is gearing up to handle these big picture issues. “I think the MSOs are starting to think through what the implications are for adaptive streaming and OTT. I would not go so far as to say that an ecosystem is forming,” Yarborough says. “I have heard they [understand they] are going to have to embrace this and manage this to own the customer.”
RGB’s Salo agrees. “The key takeaway is that adaptive streaming is new but is well beyond the hype cycle … Interesting things will happen in the next 12 to 18 months.”
Carl Weinschenk is the features editor at BGR. Reach him at email@example.com.